MCP Registry
netra-mcp, the stdio MCP server for AI agents, is listed in the
MCP Registry as io.github.zyvorai/netra (published
with v0.30.1). Look up the live entry:
https://registry.modelcontextprotocol.io/v0.1/servers/io.github.zyvorai%2Fnetra/versions/latest
| Registry name | io.github.zyvorai/netra |
| Image | ghcr.io/zyvorai/netra-mcp:0.32.0 (multi-arch, cosign-signed) |
| Transport | stdio |
Client configuration
Any stdio MCP client can run the image directly:
{
"mcpServers": {
"netra": {
"command": "docker",
"args": ["run", "-i", "--rm",
"-e", "NETRA_URL", "-e", "NETRA_API_KEY",
"ghcr.io/zyvorai/netra-mcp:0.32.0"],
"env": { "NETRA_URL": "https://netra.example:30870", "NETRA_API_KEY": "..." }
}
}
}
| Variable | Default | Purpose |
|---|---|---|
NETRA_URL | https://127.0.0.1:30870 | Base URL of the Netra controller |
NETRA_API_KEY | unset | Bearer token matching the controller's NETRA_API_KEY |
NETRA_MCP_ALLOW_MUTATIONS | false | Registers the opt-in mutating tools when true |
NETRA_MCP_ACTOR | mcp:hermes | Recorded as X-Netra-Actor, so agent actions are distinct in the audit log |
Use it
You need a running Netra controller and an API key; the MCP server only bridges to the controller's HTTP API.
Claude Code
claude mcp add netra \
-e NETRA_URL=https://<controller>:30870 \
-e NETRA_API_KEY=<key> \
-- docker run -i --rm -e NETRA_URL -e NETRA_API_KEY ghcr.io/zyvorai/netra-mcp:0.32.0
Claude Desktop or any stdio client: use the JSON above. Restart the client and the netra_*
tools appear.
Then ask in plain language, for example:
- "Why are
payments-apiconnections timing out?" - "Which kernel function is dropping traffic from namespace
prod?" - "Show TCP retransmission anomalies."
- "What changed in the network path in the last 30 minutes?"
- "Draft a containment rule for this IP, but don't apply it."
Notes:
- Inside the container
127.0.0.1is the container itself: use the node's address, or--network host. - For a lab controller with a self-signed certificate set
NETRA_TLS_INSECURE=true. - Set
NETRA_MCP_ACTORto something that identifies you or the agent; it is recorded in the audit log.
Quick check without a client (works even with no controller; real tool calls need NETRA_URL and
the key):
printf '%s\n' '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{}}' \
'{"jsonrpc":"2.0","id":2,"method":"tools/list"}' \
| docker run -i --rm -e NETRA_URL -e NETRA_API_KEY ghcr.io/zyvorai/netra-mcp:0.32.0
Read tools and mutating tools
The server exposes 215 tools: 155 read tools and 60 opt-in mutating tools. With
NETRA_MCP_ALLOW_MUTATIONS unset or false (the default), only read tools are registered; the
mutating tool names do not exist in the running process.
Enabling mutations changes nothing else about the safety model:
- Policy apply still needs a plan token and a risk confirmation.
- Enforcement stays lease-bounded and fails open to observe.
Details, the full tool list and other client setups are in the MCP integration guide.